mexMonteCarlo: Simulation from dependence models

In texmex: Statistical Modelling of Extreme Values

Simulation from dependence models

Description

Simulate Monte Carlo sample from a collection of fitted conditional dependence models.

Usage

mexMonteCarlo(nSample,mexList,mult=10)

Arguments

nSample	Required sample size.
mexList	List of fitted dependence models (returned by mexAll).
mult	Integer specifying what multiple of the total number of points should be generated for rejection sample

Details

Generates a Monte Carlo sample of the required size from a collection of conditional multivariate extreme values model of Heffernan and Tawn, 2004. For each marginal variable, the model that conditions on that margin is used to simulate values in the part of the sample space for which that margin is the largest of all marignal variables (measured on a quantile scale).

Value

A list with the following components:

nR	For each margin, number of original Monte Carlo points replaced by points generated under the corresponding conditional model.
MCsample	Matrix contiaining the Monte Carlo sample, dimension nSample by dimension of original dataset.
whichMax	Vector of indices indicating which variable is largest (on the quantile scale)
whichMaxAboveThresh	Logical vector indicating which of the variables identified by whichMax are additionally above the corresponding threshold for dependence estimation.

Author(s)

Harry Southworth, Janet E. Heffernan

References

J. E. Heffernan and J. A. Tawn, A conditional approach for multivariate extreme values, Journal of the Royal Statistical society B, 66, 497 – 546, 2004

Examples

  mAll <- mexAll(winter,mqu=0.7,dqu=c(0.7,0.7,0.7,0.7,0.7))
  mexMC <- mexMonteCarlo(5000,mAll)
  pairs(mexMC$MCsample)

texmex documentation built on June 22, 2024, 12:26 p.m.